Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

Contents lists available at ScienceDirect

Transportation Research Interdisciplinary Perspectives

journal homepage: www.elsevier.com/locate/trip

Integration of ridesharing with public transport in rural : Practice and outcomes ⇑ Vu Thi Thao a, , Sebastian Imhof a,b, Widar von Arx a a Lucerne University of Applied Sciences and Arts, Competence Center for Mobility, Rösslimatte 48, P.O. Box 2960, CH-6002 Lucerne, Switzerland b Institute of Geography, Center for Regional Economic Development, University of Bern, Hallerstrasse 12, 3012 Bern, Switzerland

ARTICLE INFO ABSTRACT

Keywords: Using advanced technologies, ridesharing has transformed transportation in many cities. The effective integra- Ridesharing tion of ridesharing into public transport systems can provide an opportunity for public transport authorities to Public transport solve the common problem of rural transport, namely poor accessibility due to the dispersal of the population. Integration However, literature on the integration of ridesharing with public transport is rather scant. At the same time, Cooperation while ridesharing has been extended to rural communities, so far this has attracted little research. To fill in Rural these gaps, this paper seeks to explore the practice and outcomes of integrating ridesharing with public trans- Acceptance port. The analysis is based on a case study of a community‐based ridesharing scheme in four rural municipal- ities in central Switzerland. Using a mixed methods approach, the key findings show the public transport authority designing the integration with a goal to complementing public transport with ridesharing. To achieve this goal, three crucial approaches were adopted: the provision of information on the ridesharing scheme, the integration of ridesharing notice boards into the local bus stop and regional railway station networks; and the price strategies as a means to signify a contractual relationship between rider and driver and to reduce a poten- tial competition. Even though ridesharing use is still only moderate, the authorities are satisfied with the out- come and do not consider cost‐effectiveness to be important.

Introduction et al., 2018), while others claim the opposite (e.g. Hall et al., 2018; Feigon and Murphy, 2016; Chan and Shaheen, 2012). Although ridesharing is far from being a new mode of transport, However, these studies have primarily focused on cities and major with the appeara + nce of mobile technology and real‐time informa- ridesharing companies. To date, little research has been conducted in a tion (Chan and Shaheen, 2012; Henao and Marshall, 2018) it has chan- rural context, even though ridesharing has been extended to rural ged the transportation market in many cities, especially in providing a communities (Hofmann and Daskalakis, 2019; Joseph, 2018). It is reliable and affordable service in neglected city areas (Hall et al., therefore of great interest to explore alternative ridesharing models 2018). Along with this significant gain, ridesharing also creates chal- in rural areas. lenges for the transport authorities in terms of demand management. Public transport in rural areas worldwide faces a common problem On the one hand, we have witnessed strong protests by traditional of poor accessibility due to the dispersal of populations (Nutley, 2003; taxi‐drivers against ridesharing companies like Uber on the basis that Petersen, 2016). Although Switzerland, like other countries in western ‘uberization’ threatens their livelihoods (Malalgoda and Lim, 2019; Europe, has higher rural densities (FSO, 2014), its public transport ser- Srinivasan, 2019). Alongside these protests there are also disputes over vices in rural areas are still characterized by low frequencies, limited authorization to operate (Contreras and Paz, 2018). On the other hand, operating hours and indirect routes (Petersen, 2012). Public transport empirical studies of the impacts of ridesharing provide contradictory users face the first/last mile problem (Reck and Axhausen, 2019). Poor findings. Some studies report that ridesharing attracts riders away public transport services also cause a high dependence on private cars. from public transport (Franco et al., 2019) and increases vehicle miles Nevertheless, around 22% of the Swiss rural population do not own a travelled (Henao and Marshall, 2018) and traffic congestion (Hall car (FSO, 2017), leaving the travel needs of the non‐car population, as

⇑ Corresponding author. E-mail address: [email protected] (V.T. Thao).

https://doi.org/10.1016/j.trip.2021.100340 Received 5 December 2020; Revised 5 March 2021; Accepted 7 March 2021

2590-1982/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340 well as the elderly and young people,1 unmet. Meanwhile public trans- taxi. In a ridesharing system, the driver and the passengers share a port authorities are increasingly being pushed to improve the efficiency common route (origin and/or destination); conversely ridehailing dri- and effectiveness of how public transport is organized. They are already vers typically do not have the same destination as their passengers required to ensure the provision of public transport to meet a range of (Chan and Shaheen, 2012; Clewlow and Mishra, 2017). social policy goals, such as social equality and sustainable mobility Based on modes of operation and participants’ relations, Chan and (Chen et al., 2017; Docherty et al., 2018). The effective integration of Shaheen (2012) classify a ridesharing scheme into three subgroups. ridesharing into the public transport system can provide an opportunity The ‘acquaintance‐based’ carpool is formed by families, friends and for the public transport authorities to improve accessibility and to coworkers. The ‘organization‐based’ carpool is arranged around a for- increase public transport usage in rural areas by addressing the first/last mal institution: to use this service, formal membership or a visiting mile problem (Stiglic et al., 2018), and managing ridesharing as a com- organization’s website is required. The ‘ad‐hoc’ form of ridesharing plement to public transport as seen later in the empirical part of this cur- involves casual carpooling in which the driver and the passengers rent paper. Yet the literature on the integration of ridesharing and public do not know each other. Membership is not required. The shared transport is rather scant. Moreover, to persuade the authorities to pursue rides are organized by various mechanisms such as self‐ this solution, future research should document successful practices organization, notice boards/meeting points and computerized ride- through case studies (Murray et al., 2012). matching systems. Because of the characteristics of the case study, Against this backdrop, the present paper seeks to explore the prac- this current paper only focuses on the integration of ridesharing with tice and outcomes of integrating ridesharing with public transport in a public transport. rural context. We first pay special attention to the ways in which key Rural areas lag behind urban areas on digitalization due to their stakeholders cooperate with each other in designing and implementing physical remoteness and low population densities (Salemink et al., the integration process. We then look at the outcome of the integration 2017). The available ITC infrastructure in rural regions influences by means of a descriptive analysis. Because identifying suitable geo- the technologies adopted by ridesharing schemes. In addition, unlike graphical areas plays a crucial role in designing the integration of urban areas, where ridesharing services are more individual‐oriented ridesharing with public transport, we also examine the impact of fac- and profit‐oriented, ridesharing models in rural areas are often char- tors in the built environment on ridesharing use. To do so, we adopt acterized by ‘community participation, individual altruism and col- a mixed methods approach by combining qualitative interviews with laborative initiatives’ (Joseph, 2018, p. 3). A study conducted in data on level of demand and geographical information. The analysis rural Scotland also reports that strong local social networks have is based on a case study of a community‐based ridesharing scheme, positive impacts on the feasibility of ridesharing (Gray et al., 2001). Taxito, in four rural municipalities of central Switzerland. Our findings Previous research has shown that the built environment has certain highlight that filling a gap in public transport and mitigating outmigra- impacts on travel behavior (Thao and Ohnmacht, 2019). However, the tion to cities are the most important reasons for integrating ridesharing relationship between the built environment and ridesharing remains with public transport. Even though ridesharing use is at present mod- under‐researched (Yu and Peng, 2019). While empirical evidence from est in amount, the authorities are satisfied with the outcome of this a few existing studies confirms the impacts of the built environment on scheme and do not consider cost‐effectiveness to be important. Instead ridesharing behavior, findings on the magnitude of the impact proved they are concerned much more with how to design the integration contradictory. Conducting a study of ridesharing in Texas, Pu and Peng such that ridesharing should complement and enhance the use of pub- (2019) claim that there are strong associations between the built envi- lic transport. Our paper also enriches knowledge of the associations ronment (density, land use, infrastructure and transit accessibility) and between ridesharing and the built environment, another underex- the demand for ridesharing. However, a study of ridesharing in Los plored research area (Yu and Peng, 2019). Angeles conducted by Brown (2020) indicates relatively weak impacts The remainder of the paper is organized as follows. Section 2 of built environment factors on ridesharing, in comparison to neigh- briefly introduces the ridesharing terminology, the classification of borhood socioeconomic characteristics. These two studies were con- ridesharing models, the impacts of the built environment on rideshar- ducted in a city context. In rural settings, there have been no studies ing, and the integration of ridesharing with public transport. Section 3 of this issue to date. In the empirical section of the paper, we will pro- describes the context for the study, followed by a presentation of the vide initial findings from our case study of four rural municipalities in data and methods used in Section 4. Section 5 discusses findings about central Switzerland. the actual integration process and its outcomes. Section 6 assesses the In practice, several ridesharing models have been developed as implications for policy‐makers in integrating ridesharing with public complementary to public transport services in rural areas. For exam- transport in a way that increases use of the latter. ple, Carlos was the first ridesharing service to be implemented in a non‐urban context in Switzerland. It was simply incorporated into local and regional public transport services by installing call boxes Literature review around bus stops and railway stations. The project was ended after the pilot phase (2002–2005) due to the long waiting times, weak com- As mentioned previously, the integration of ridesharing with public munication and low demand (Hofmann and Daskalakis, 2019). The transport is still under‐researched. This literature review therefore Mobilfalt ridesharing scheme, implemented in 2013 in north Hesse, adopted an integrative review method to provide a knowledge base Germany, has achieved stronger integration because it is combined and to present insights from different integration models. We begin with the routes and timetables of local and regional public transport with the terminological distinction between ridesharing and ridehail- services. It bundles not only private cars but also taxis with public ing. Both ridesharing and ridehailing refer to carpools (or vanpools) transport modes (Benz and Kepper, 2019). The Garantiert Mobil in which two or more individual travelers share a journey in a private ridesharing scheme implemented in the south of Hesse goes a step fur- automobile. However, ridesharing differs from ridehailing in its finan- ther in being fully integrated into local and regional public transport cial goals and types of travel route. Ridesharing is non‐profit because services (routes, timetables and tariffs). Besides private cars and taxi, the payment only covers part of the driver’s costs. Conversely, ridehail- it bundles rental cars with modes of public transport (Krämer and ing by, for instance, Uber, Lyft or Grab is a variation on the for‐profit Weiss, 2019). Other common characteristic between Mobilfalt and Garantiert is the active involvement of local transport agencies and the municipal authorities, which finance a large share of the operating 1 The number of Swiss people applying for driving licenses has declined in recent years. In 2017, this figure fell by 2% for the whole population. The strongest reduction (3%) was costs, thus reducing fares for riders (Benz and Kepper, 2019; Krämer in the young population group (ages 18–24) (ASTRA, 2018). and Weiss, 2019).

2 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

Fig. 1. The area of study.

A survey of public transport agencies and ridesharing operators in formed a working group to share practical experiences and to stress the United States conducted by Murray et al. (2012) is one of a few the potential of this mobility solution to the Federal Office of Transport. studies devoted to the integration of ridesharing with public transport. The case study for this paper is Taxito, the successor to the Carlos According to these authors, bridging the gap in transit services in areas ridesharing scheme mentioned earlier. A key feature of Taxito with sparse populations and addressing customer demand are the most ridesharing is that it aims to complement public transport and operates important reasons for ridesharing operators and public transport agen- in areas with no or limited public transport services. Taxito has the cies to cooperate. To evaluate the outcome of the integration, rideshar- characteristics of a community‐based, organization‐based and ad‐hoc ing services are commonly used by both ridesharing and public ridesharing scheme. It requires no membership registration for either transport agencies. For the latter, cost‐effectiveness is reported as riders or drivers. However, if the driver wants to receive a free high- being less important. This suggests a reason why few public transport way charge sticker (worth CHF 40) and payments (CHF 1 per ride), agencies have considered substituting ridesharing for a fixed route as a he or she can register at the operator’s website. To use Taxito rideshar- cost‐saving measure. ing, a rider walks to the nearest Taxito notice board and sends the Tax- ito platform a destination code via mobile phone. The destination name will be lit up on the notice board. A driver sees the destination The study context and can spontaneously decide whether or not to take the rider. To ensure security, the rider must send the vehicle’s registration number In the past few years, several pilot projects (e.g. Taxito, Ebuxi/my- to the Taxito platform. The rider must pay CHF 2 per ride, which is buxi, Kollibri, sowiduu) promoting the integration of on‐demand directly charged by a mobile network operator. The price can cover mobility with public transport have emerged in Switzerland. These only a small proportion of the operating costs. projects have been implemented in urban, peri‐urban and rural areas We chose Taxito for our investigation because it is the most wide- and are varied in their technological characteristics and product spread ridesharing scheme to operate in low population density areas design, but all seek to achieve synergies between on‐demand mobility of Switzerland. This suggests that Taxito has certain advantages that and public transport. On‐demand bus and ridesharing are designed to other ridesharing schemes do not. In Switzerland, Taxito is currently combine with local bus routes and/or to connect with regional railway operating in , , Zell and (Canton stations. It is therefore not surprising to see the strong involvement of Lucerne), Maladers (Chur), Hé Léman (Geneva) and Emmental‐ public transport operators like the Swiss Federal Railways (SBB), Entlebuch (Bern and Lucerne). We selected Taxito’s activities in the PostAuto,2 and cantonal and regional public transport associations. four rural municipalities of Luthern, Willisau, Zell and Grossdietwil Besides their duty to meet customer demand, by becoming involved in for investigation here. The first three municipalities were the first these projects, public transport operators can increase their customer where Taxito began operating in June 2015. This timeframe can be base and protect their market positions. Project stakeholders have considered long enough to be able to evaluate the outcome of the inte- gration of Taxito ridesharing with local public transport. Furthermore, 2 PostAuto provides rural and regional bus services throughout Switzerland. other operational areas cover either part of some agglomeration or are

3 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340 still in the pilot phase. However, we included Grossdietwil in our of the road network and therefore does not form circular isochrones. study, even though Taxito ridesharing only started in this municipality For all thirteen catchment areas, isochrones of 500 and 1000 m were on 30 December 2019. This is because we are interested in knowing calculated. The geographical distances used were chosen because they why Grossdietwil decided to participate in the Taxito ridesharing pro- are equivalent to the accessibility of the public transport connections ject at a much later date, even though it is located in the same neigh- in rural areas. The accessibility is defined by a distance to the transport borhood and faces similar problems with public transport. stop from the household location (FOSD, 2020). In order to investigate As can be seen in Fig. 1, the municipalities all have low population the proportion of the population with access to the notice boards densities, ranging from 20 person/km2 in Luthern (and Luthern Bad) within a walking distance of 500 and 1000 m, the population living to 180 person/km2 in Willisau. Due to low demand, the quality of pub- within the areas defined by each isochrones was determined by clip- lic transport services is rather poor. There is a regional train connec- ping the isochrones with the dataset of Swiss statistics for the Swiss tion serving the traffic corridor Willisau – Gettnau – Zell – , population (Statpop) (FSO, 2018a). This provides the population num- but it only runs once an hour and every half hour at rush hours. A ber on a hectare grid. This step was repeated for employees working bus line serves the traffic corridor Zell – Hüswil – Luthern – Luthern Bad, but it runs much less frequently, only five times on weekdays and four times at weekends. Another bus line serves the traffic corridor Table 1 Hüswil – Zell – Grossdietwil with a direct connection seven times on Taxito trip data Taxito data, n = 3562 trips. weekdays and twice at weekends. Variable Description Like other rural areas, the share of individual motorized trafficin fi Date total traf c volumes is high in these communities. At the same time, Time of day the average vehicle occupancy rate is rather low, at 1.37 person/car. Place of Taxito’s notice boards These factors provide favorable conditions for the implementation of departure a ridesharing service. Destination Destination of the trip. No information on the actual place of arrival With its very sparse population density and therefore very poor Waiting time The time from the rider requesting the service to the time public transport services, Luthern was the municipality where the inte- confirming the successful match. The real waiting time could be gration of Taxito ridesharing with public transport was initiated. Luth- shorter because the rider might not send a message immediately ern has two notice boards, instead of one notice board as in the other after being picked up three municipalities. From each notice board, riders can request a trip Registered rider Yes/No. This information was included in the dataset recorded Registered from 1 June 2015 to 9 January 2019 to different destinations in the neighborhood. driver Confirmation A short message sent by the rider to confirm the successful message match. It might also contain the number plate of the vehicle Data and methods picking up the rider. The authors of this paper received only the first two letters of the number plates indicating the cantons in In this study, we have used a mixed methods approach. To explore which the vehicles are registered. This information was included the ways in which the public transport authorities cooperates with the in the dataset recorded from 10 January 2019 to 13 June 2019 Taxito company in implementing the integration of ridesharing with To investigate the impact of aspects of the built environment on ridesharing ‐ public transport, three in depth interviews with the founder and use, we combined the demand dataset in Table 1 with a dataset for the built 3 CEO of Taxito company, the Lucerne Transport Association and a environment. Built environment data and sources are summarized in Table 2. member of the municipal council of Luthern were conducted. The inter- views were tape‐recorded and lasted between one and two hours. These interviewees were selected because they were responsible for the design Table 2 and implementation of the ridesharing service in the study area. The Data on the built environment. interviews covered the following topics: cooperation goals and activities, stakeholder involvement in the implementation process, cost‐ Variable Description Source effectiveness and challenges. In addition, secondary data like technical Population Number of people living in catchment (FSO, 2018a) and evaluation reports on the pilot project were also collected. areas of 500 m and 1000 m around To assess the outcome of the integration of Taxito’s ridesharing municipality/settlement center scheme with public transport, we used an anonymized dataset of Employees Number of employees working in (FSO, 2018b) 4 catchment areas of 500 m and 1000 m 3562 trips recorded from 1 June 2015 to 16 March 2020. For each around municipality/settlement center trip, Taxito and the Lucerne Transport Association provided eight vari- Point of Classification of amenities in the (Openrouteservice, ables, as listed in Table 1 below. interest neighborhood that can influence the use of 2020) The accessibility of the notice boards and destinations was ana- (POI) ridesharing services, including - Company and small businesses (e.g. lyzed in order to investigate the impact of population and employment retail, supermarket, health service) densities on the use of Taxito ridesharing. We plotted the catchment - Restaurants and entertainment areas for each of the five notice boards and the eight destinations by - Residential areas (e.g. camping site, using the Open Routes Service ‘ORS Tools’ plug‐in in QGIS. Catchment hotel, retirement home) areas were defined as polygonal areas with their centroid placed on - Public organizations (e.g. school, church, ’ ‐ police station) Taxito s notice boards and at the destinations on the center most point - Tourist attractions (e.g. pilgrimage of a transport axis within a municipality or settlement, with all their destination, tourist information, animal vertices placed on connected transport axes (roads, streets, paths). zoo) The catchment area is determined by fixed geographical distances - Infrastructure and other (e.g. parking, gas station, fireplace) (500 or 1000 m) along all transport axes from their centroid to their Regional Location of regional railway stations (SBB, 2020) vertices. The shape of the catchment areas depends on the structure railway network Daily Individual traffic model with average daily (FOSD, 2017) 3 The interviewee’s identity was concealed at the interviewee’s request. motorized traffic 4 From 17 March 2020 to 24 September 2020, the Taxito ridesharing scheme had to traffic suspend its services due to prevention measures connected with Covid‐19 pandemic. This volume was because the local authorities considered Taxito to be a public transport service.

4 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340 within the catchment area defined by the isochrones in order to deter- who has access to a cell phone can use Taxito ridesharing, and it can be mine the number of employees with access within a walking distance accessed by low‐income population groups such as students. It costs of 500 or 1000 m by using the dataset of Swiss statistics for the struc- just 75 pennies more than a single journey on a bus ride, so that Taxito ture of enterprises (Statent) (FSO, 2018b). This provides the number of ridesharing does not compete with public transport providers in terms employees on a hectare grid. The same analytical procedure was car- of price. However, the price of a journey by ridesharing must be much ried out for the destinations. cheaper than a taxi ride, which is CHF 3 per kilometer. Moreover, Tax- To explore the impact of land‐use diversity on ridesharing, we first ito destination display boards by means of lights increases the ride’s counted the points of interest (POIs) in the immediate vicinity of each appeal to private car drivers, thus helping to reduce waiting times. notice board and destination by using the POI application program- Another added value of Taxito ridesharing is that can be available ming interface (API) Openrouteservice (2020). Adapting the cate- 24/7 by exploiting the spare capacities of individual motorized traffic. gories of POIs by Kong, Zhang and Zhao (2020) for rural areas, we assigned each POI to one of the six categories shown in Table 2.We Identification of a suitable neighborhood used the POI categories to calculate land‐use diversity, represented To identify a suitable neighborhood in rural areas, VVL adopted a by the Shannon entropy (Shannon, 1948): phased adoption approach. The agency considered the Taxito rideshar- ing scheme to be a new mobility solution, a kind of innovation. Using n ¼∑ H pjlognpj the phased adoption approach, the public transport authority expects j decisions to be made in a transparent manner, the experience gained then being taken into account in taking further decisions. It first con- where H represents the entropy with a range from 0 to 1; pj is the pro- ducted an analysis to evaluate the suitability of the Taxito scheme by portion of each category of POI (j); and n is the number of categories, taking into consideration the quality of its transit accessibility, popula- here eight categories. The higher the entropy, the higher the land‐use tion densities and daily individual motorized traffic volumes (VVL, diversity. 2014). Based on the results of this analysis, Luthern, Willisau and Zell In Taxito’s ridesharing concept, notice boards can be considered to were selected for the pilot phase. The goal of the pilot phase was to test be the places where trips originate. We plotted the number of trips the system and probe acceptance among the local population, the local recorded between a notice board and its possible destinations. In this municipalities and the regional development agency, rather than gain- step, we added the regional railway locations (SBB, 2020) in the ing large numbers of riders. After receiving positive feedback from the municipal neighborhood in order to explore the possible impact of municipal councils and the regional development agency, VVL decided regional railway locations on ridesharing use. In addition, daily motor- to start the pilot phase. ized traffic volumes and the directions of traffic were taken into con- sideration (FOSD, 2017) to determine the relationship, if any, Location of notice boards and destinations between the number of trips departing from each notice board and The most important task in the pilot phase was to decide the loca- daily motorized traffic volumes and directions. tion of the notice boards. There is general agreement among VVL, the municipalities and Taxito that the notice boards should be integrated Results into the local public transport network. The location must be identified on the basis of the quality of transit accessibility, that is, a poor public The practice of the integration process transport connection or none at all. A high daily individual motorized traffic frequency is another important criterion for location choice. The Goals of the integration location must allow private car drivers to follow the road transport The actors involved in the design, implementation and operation of regulations, for instance, no prohibition on stopping and the necessary the Taxito ridesharing scheme are the Lucerne Transport Association distance to crossroads. Ideally, individual motorized traffic should be ’ (VVL), the Taxito company and the local municipalities. VVL s roles sorted by direction of travel and connected to destinations requested fi in general include planning and nancing public transport in the can- by Taxito riders, thus contributing to more successful ‘matching’. ton of Lucerne. In this particular ridesharing project, VVL is a contract- At the same time, the location should not disturb the public trans- ’ ing authority that sets goals and provides funds. VVL s goal is to port service. The location should also have pedestrian crossings, so improve public transport services in local municipalities. The agency that riders have easy access to the notice boards. Furthermore, the requires a design for the integration in which ridesharing is treated location must guarantee certain security for riders (see Picture 1). Dur- as an integral part of public transport. The ridesharing service should ing the pilot phase, the location of the notice boards must be adjusted complement public transport in terms of both frequency and routes. to end‐user demand. In principle, there should be one notice board for fi There is no clearly de ned goal for how many trips should be under- every municipality. However, if demand increases, additional notice taken by ridesharing per year. It is important to stress that the public boards can be installed. As a result, except in Willisau, where the transport authority is not attempting to reduce the ownership of pri- notice board is located around 200 m from the railway station, in all vate cars by introducing the ridesharing service. In fact, it relies on pri- other municipalities, they are located at the bus stop. Thanks to the vate cars to achieve its goals. rigorous criteria for the choice of notice board locations, until now In addition, the main reason for improving public transport in local no damage, accidents or assaults have been reported. ‐ municipalities is that the latter expect Taxito ridesharing to reduce out To install the notice boards, budget applications and an approval migration to cities by making village life more attractive to the local request had to be submitted to the municipalities to reconstruct the population, especially young people. They also hope that the rideshar- road space. Approval for the installation of the illuminated signboard ing service can promote local tourism. Day tourists are therefore con- had to be obtained from the cantonal road traffic authorities. At this fi sidered to be another bene ciary group, apart from local people. stage, the municipalities were closely involved in the decision‐ making process because they had good knowledge of local conditions Selection of a suitable on-demand mobility service and understood the needs of their communities. They also provided As mentioned earlier, there are several ridesharing models in infrastructure and budgets for the installation of the notice boards. Switzerland currently. VVL has also evaluated other transport service The Taxito company supported the municipalities in determining the models like regular buses, on‐demand buses, collective taxis location of the notice boards and destinations. The company is in (Anrufsammeltaxi), etc. VVL chose Taxito ridesharing because it offers charge of maintenance and the provision of a replacement in the event a low threshold for both riders and drivers using the service. Any rider of a defect.

5 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

Picture 1. Taxito board (Source: Keystone/Urs Flueeler).

Each Taxito point can connect to several destinations (see Fig. 7 It also pays for renting four Taxito notice boards (VVL, 2014). The below). Locations of destination are decided based on the following municipality of Grossdietwil pays for the notice board itself. The principles: using Taxito ridesharing as a means to overcome the municipality decided to join the Taxito ridesharing network having first/last mile problem by transporting villagers in the municipalities observed the good experience with it in the neighborhood. to/from regional public transport hubs and as a means of direct con- The municipalities finance the installation of notice boards and nections with other villages in valleys to avoid indirect connections provide solar panels for them. The Taxito company makes its revenue associated with local bus services (for instance Luthern Taxito point by renting out the notice boards and by charging 0.1 CHF per ride. The – Hergiswil). Destinations are therefore located near regional railway remainder of a charge per ride goes to a telecom company and the reg- stations and/or in villages in the valleys. istered driver or, if the charge is waived, to VVL. This suggests that the Taxito company makes a very low profit, but it still provides the Acceptance of local people ridesharing service because it sees itself as a social entrepreneur pro- The success of the Taxito ridesharing scheme depends also strongly moting an environmentally friendly mobility solution that contributes on the goodwill of individual car‐drivers and the local people’s willing- to the attractiveness of village life. ness to use the service. The stakeholders therefore see local people as To find out whether the Taxito ridesharing solution is a cost‐ an important factor in the success of the integration of ridesharing effective solution or not, we calculated the total costs of the entire with public transport. They decided that the municipalities should take ridesharing scheme and the cost per ride, based on the costs provided responsibility for promoting public acceptance of the scheme. As ‘in- by the interviewees. The rent for one notice board is 5988 CHF in the siders’, they have direct access to the local population. However, as first year and 3950 CHF for each of the following two years. The instal- the municipalities had little experience in staging communication lation of one notice board costs 4450 CHF. In total, in the first three and awareness campaigns, Taxito worked closely with the municipal- years and for all five notice boards, the entire Taxito ridesharing ser- ities in this activity. It provided the municipalities with communica- vice will cost 91,690 CHF. The costs of planning and obtaining tion equipment, for example, and supported them in organizing and approval of the locations is not included because it is too complicated holding marketing and communication events for the launch of the to estimate the human resource costs of local authorities. pilot phase. Fig. 2 shows the development of the costs per ride and the total A series of intensive communication and marketing activities were costs of the service. In the first three years, the ridesharing service is carried out to raise awareness and increase acceptance of Taxito ridesharing. Two weeks before launching the pilot phase, a flyer intro- 40 29500 duced the goal of Taxito’s scheme and explaining how it works was 35 29000 sent to each household in the municipalities. Awareness events to 30 28500 introduce the concept and functions of ridesharing were held at local 28000 25 schools. Meanwhile, several media press events were organized at 27500 the invitation of representatives of local media agencies, like newspa- 20 27000 pers and television and radio programs before and at the launch of the 15 pilot phase, at the opening of the first notice board and before the end- 26500

10 Total cost (CHF) ing of the pilot phase. The launch of the pilot phase even received Cost per ride (CHF) 26000 attention from national media agents. Meanwhile, the testimonies of 5 25500 local people as both drivers and riders were collected and used in com- 0 25000 munications and advertising concerning the ridesharing. 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 Number of trips

Participation in costs Cost per ride (CHF) Total cost (CHF) As the contracting authority, VVL bore the costs of the initial anal- ysis, the pilot phase, and the marketing and communication activities. Fig. 2. Cost of Taxito ridesharing service.

6 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340 shown as being used by an average of 708 passengers a year, without Satisfaction with ridesharing the Grossdietwill notice board. By adding 180 passengers who would Despite the above challenges and weaknesses, VVL is satisfied with use the Grossdietwill notice board, we arrive at a total of 888 passen- the outcome. The contracting authority views a low but constant num- gers using the ridesharing service per year. The total cost (fixed costs ber of trips, given the sparsely populated area and low level of market- minus fare) per ride is around 32 CHF. That is a high amount. How- ing effort, as positive. Even though the municipalities agree on this, ever, the total cost would be reduced over time if the number of pas- there is nonetheless some skepticism. One municipal councilor sengers remained stable or increased. Nevertheless, the total cost of reported that he has never seen anyone at a bus stop and accordingly the Taxito ridesharing service is still much lower than a single regular has never taken anyone with him. There is even a certain distrust of bus line that runs five times a day and would cost 1 million CHF per the statistics because, ‘if it were needed, you would have seen people?’ year to run in rural areas, with just 30% of the cost being covered As our descriptive results below show, local people do use the by the fare. ridesharing service, but only at a modest level. Tourists seem to use Collecting the fare of 2 CHF per ride is actually expensive due to the the service more often. Taxito ridesharing can probably be seen as high roaming cost charged by telecom providers. However, VVL does an enhancement of tourist provision in the municipalities. not want to provide a free‐of‐charge ridesharing service. Beside com- VVL is also satisfied that the Taxito ridesharing service does not petition avoidance, appreciation is another reason to charge a fare: compete with local public transport once it has been in operation for the fact that the ride costs the rider something establishes a contrac- longer. Instead, it increases the modal split in local bus services in tual relationship between the driver and the rider. Ridesharing trips the third year of the ridesharing operation, though at a very modest should not be perceived as ‘charity’ work, according to the contracting rate (VVL, 2017). As the notice boards are at bus stops, it sometimes authority. happens that riders take the bus while waiting to be picked up by the ridesharing service if a bus arrives beforehand.

The use of Taxito ridesharing Challenges and weaknesses Optimization of the ridesharing service in regular operation has During the period from June 2015 to March 2020, 3562 trips were proved to be a big challenge. For example, the location of the notice made using Taxito ridesharing. Of 2215 trips travelled from 2015 to board in Willisau is not considered suitable because potential riders 2018, only 99 were made by registered riders and only 36 by regis- have to walk some distance away from the center of the municipality tered drivers. Of 302 recorded trips containing confirmation messages, to get to the notice board. However, it has not been moved somewhere 85 trips only had information using the words ‘OK’, ‘good’, ‘go’ or ‘car’. better, mainly because no clear responsibly for doing so was estab- The remaining 217 trips were undertaken by local drivers (61%), dri- lished after the pilot phase. Furthermore, the number of notice boards vers from neighboring cantons (17%), drivers from foreign countries are viewed by the Taxito company as requiring expansion. The intro- (15%) and buses (7%). duction of more notice boards would increase ridesharing use. The As can be seen in Fig. 3, the number of trips fluctuates overtime, authorities agree on this point, but they have other priorities than pur- reaching a peak in the second year. This could be due to the effects suing this goal. One reason for the reluctance to expand is that of the marketing campaign at the end of the pilot phase in which ridesharing could become a competitor to local public transport awareness of Taxito ridesharing was widespread in the local munici- operators. palities and nearby neighborhoods. For the last ten months of opera- The Taxito company itself speaks of a relative high level of accep- tions (June 2019–March 2020), seven hundred trips were made, tance and sensitization of the local population to the relevance of roughly as many trips as in each of the two previous years. This ridesharing. However, it is difficult to maintain the acceptance and increase in demand may suggest that Taxito ridesharing has become willingness of drivers to take part. The basic belief that ‘Taxito brings an established name and aspect of local public transport services. something to our region – we participate’ (Taxito founder) must be On average, riders have to wait less than four minutes to be trans- regularly communicated to the public. While Taxito is relatively well ported. This is a very short waiting time when compared to the low fre- known in very remote areas in those municipalities without public quency of public transport services. The waiting time is expected to transport services, drivers in regional centers (where regional railway decline still further in the first four years but to increase again in year stations are located) hardly know anything about Taxito’s ridesharing five. There are some trips for which riders had to wait more than hour, scheme. Since the completion of the pilot phase, no institution has usually late at night or early in the morning. been made responsible for marketing the scheme. There is no service Fig. 4 shows that most trips are made between 15:00 and 19:00 and mandate for this, and accordingly no responsibility and budget. between 21:00 and 23:00. There is no great demand for ridesharing Nonetheless, except in the Luthern municipality, which is affected by during the morning rush hour. This is because local buses have more outmigration, awareness activity is still somehow carried out at a cer- tain level. Since Taxito is not integrated into public transport’s central com- 4.5 1000 munication channels (timetable and ticketing channels), there is an acute danger of the service being neglected. During the Covid19 lock- 4 900 down, Taxito was banned from operating. However, this information 800 was not shown on the notice boards. Riders did not receive any infor- 3.5 mation via mobile phones. Only by visiting the website of the Taxito 700 3 company could they access to this information. Therefore, the intervie- 600 Waing me wees agree that communication and marketing activities should be Number of trips implemented after the pilot phase in order to maintain local people’s 2.5 500 participation in Taxito’s ridesharing scheme, especially in the regional 2 400 centers. 2015-2016 2016-2017 2017-2018 2018-2019 2019-2020 While Taxito sees simplicity of use as a central and unique selling Axis Title fi point, the municipality nds it rather complicated. Because the rider Number of trips Waing me (in minutes) has to enter a message code to activate the notice board and send the vehicle’s number plate. Fig. 3. Number of trips and waiting time.

7 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

400 ends may be because of the even lower frequencies of public transport services on Saturdays and Sundays. 300

200 Impact of the built environment on Taxito ridesharing use Number of trips 100

0 Accessibility of notice boards and destinations 1 2 3 4 5 6 7 8 9 101112131415161718192021222324 The notice board in Willisau is accessible to more than 2000 people living and over 1000 people working at a walking distance of 500 m Fig. 4. Daily time travel. (see Figs. 6a and 6b). Doubling the walking distance also doubled the population and number of employees with access to the notice board. Similar results were found in Zell. In both cases, notice board 600 use is high, with 962 trips starting in Willisau and 1166 in Zell (see 500 Table 3). However, a low level of population and employee accessibil- ity to a notice board does not necessarily correlate with a low usage for 400 that notice board. Of the five notice boards, that in Luthern Bad had the lowest population and number of employees in both of its catch- 300 ment areas, but it had the second highest usage, with 1006 trips. The notice board in Luthern recorded 408 rides, less than Luthern 200 Bad, despite its larger population and number of employees with 100 access to it. Regarding user demand for travel to different destinations, despite 0 having both a high population and a high number of employees, Hut- Mon Tue Wed Thu Fri Sat Sun twil shows only a very low number of trips. A similar relationship can be found in Gettnau (see Table 3 below). This suggests that there is no Fig. 5. Distribution of trips throughout the week. direct correlation between the size of the population or the number of employees and the number of trips to a destination. The potential frequent services at rush hours. It seems that riders use Taxito demand for travel to Luthern Bad, Hübeli and Hüswil can be consid- ridesharing at night hours mainly for non‐work‐related trips. ered low, taking into account the number of possible users reached. As shown in Fig. 5, the demand for Taxito’s ridesharing scheme However, the actual level demand suggests the opposite: the numbers gradually increases throughout the week. The high demand at week- of recorded trips to these destinations are among the top five. This

Fig. 6a. Population in catchment areas of 500 and 1000 m of notice board locations and of destination locations.

8 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

Fig. 6b. Employees in catchment areas of 500 and 1000 m of notice board locations and of destination locations.

Table 3 For the destinations, a similar unclear relationship between the num- Land-use diversity at notice boards and destinations. ber of trips and the land‐use diversity index was observed. A higher land‐ Destinations Total trips from Total trips to Shannon use diversity index does not lead to a higher number of trips to a desti- notice board destination entropy H nation. For example, Willisau (Shannon entropy of H = 0.90) was found fi Grossdietwil 20 857 0.90 to be the nal destination of only 42 trips, whereas Hüswil, with a rela- Luthern 408 29 0.81 tively lower Shannon entropy (H = 0.64), is the most frequent destina- Willisau 962 42 0.90 tion in the entire Taxito network. Luthern Bad 1006 691 0.62 Zell 1166 334 0.90 – Gettnau 5 0.70 fi – 9 0.72 Daily individual motorized traf c and regional railway stations Hofstatt – 26 0.36 Fig. 7 shows two highly used traffic flows in the ridesharing net- Huttwil – 76 0.84 work in the neighborhood. The highest number of trips was observed – Breiten, Heim 80 0.58 between Zell and Grossdietwil (852 trips), followed by trips between Hergiswil – 134 0.88 Hübeli – 309 0.39 Luthern Bad and Hüswil (770 trips). Hüswil – 970 0.64 Of the total of 27 possible connections between notice boards and destinations, seventeen connections recorded fewer than fifty trips. They are therefore considered connections of low demand. Of these seventeen low demand connections, six are between two railway sta- tions. However, the presence of a regional railway station at either comparison between population and number of employees within 500 of the two ends of a connection (either near the notice board or at and 1000 m and trips ending at destinations show no clear patterns. the destination) appears to be correlated with a higher number of trips in some cases. The destination of Hüswil, which has a railway station, Land-use diversity was the most frequent destination (see Table 3). The number of trips As shown in Table 3, the notice board locations have a high diver- from Luthern Bad to Hüswil and from Zell to Grossdietwil are the sity of land uses, except in Luthern Bad (Shannon entropy H = 0.62). two most frequent connections in the ridesharing network. In both Comparing the total trips departing from each notice board, no associ- cases, the regional railway station is situated at the end of a traffic axis ation between the land‐use diversity index and the demand for Taxito going out of the local valley. ridesharing was found. A low level of land‐use diversity does not imply As can also be seen in Fig. 7, the existence of a well‐frequented low use of the notice board. route has a positive impact on the success of ride ‘matching’. If a route

9 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

Fig. 7. User demand of traffic connections between notice boards and destinations in relation to locations of regional railway stations and average daily motorized traffic volume.

has a higher daily volume of motorized traffic, it has a tendency to notice boards with regional railway stations can bring customers to have a higher number of ridesharing trips. For example, the routes railway operators, thus promoting rail use. from Luthern Bad to Hüswil and from Zell and Grossdietwil have aver- The third approach is the pricing strategies. The ridesharing ticket age daily traffic volumes of between 2500 and 5000 cars and the high- is slightly more expensive than the bus fare but much cheaper than est number of ridesharing trips, from 500 to more than 850. Likewise, taking a taxi. This pricing strategy can ensure that ridesharing does a low daily average traffic volume (below 500 cars/day) on the routes not replace public transport. Empirical evidence from the demand used by ridesharing, e.g. between Luthern and Huttwil, appears to dataset shows that the public transport agency has achieved this goal. have a low number of recorded ridesharing trips. This explains why the authority is satisfied with a relatively low but stable use of ridesharing over time. Our findings support Murray Discussion and conclusion et al.’s claim (2012) that for the public transport authorities, cost‐ effectiveness is less important in seeking integration with ridesharing. This paper has attempted to explore the practice and outcome of In the context of rural Switzerland, the goal of reducing outmigration the integration of ridesharing with public transport in a rural context. to urban areas appears to have a higher priority. Furthermore, the low The findings show the public transport authority carefully designing fare is used as a means to signify a contractual relationship between the integration with a view to complementing public transport with the rider and driver, thus contributing to the appreciation of a ridesharing. Three crucial approaches were adopted to achieve this ridesharing ride as a service rather than as a kind of charity work. goal. First, Taxito’s ridesharing scheme was developed and imple- Our findings suggest that public transport operators, particularly mented in a participatory way, with strong engagement by the munic- the railway operator, can expand their customer base thanks to the ipalities and local populations concerned. The provision of information integration of their services with ridesharing. Therefore, local bus on the scheme before and during the pilot phase was carried out in a and railway operators should actively work with the ridesharing series of intensive communication and marketing activities to raise operator in extending a network of ridesharing notice boards and awareness of local people and to encourage their participation in Tax- incorporating ridesharing into their central communication chan- ito’s ridesharing. nels, for instance provide information on ridesharing services on dis- Second, ridesharing notice boards are integrated into the local bus play boards at regional stations. The public transport authority (VVL) stop and railway station networks. This way of integrating ridesharing could consider to include these operators in the ridesharing scheme. services with public transport systems can promote the usage of public For the ridesharing operator, it should propose a budget plan for transport services. Instead of continuing waiting for being transported communication activities beyond the pilot phase so that it can com- by the ridesharing service, riders can take a bus trip if the bus arrives municate with its customers when major disturbances occur such as beforehand. Indeed, a slight increase in the ridership of local bus ser- sending text messages to its customers or regularly hold awareness vices has been observed. Further, the connection from ridesharing campaigns.

10 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

Two aspects arose on the basis of the findings of this study that Brown, A.E., 2020. Who and where rideshares? Rideshare travel and use in Los Angeles. – should deserve further investigation. One is why the use of Taxito Transp. Res. Part A Policy Pract. 136, 120 134. https://doi.org/10.1016/ j.tra.2020.04.001. ridesharing rather low despite its advantages. Does this problem lie Chan, N.D., Shaheen, S.A., 2012. Ridesharing in North America: past, present, and in the acceptance of local people? We therefore propose a hypothesis future. Transp. Rev. 32, 93–112. https://doi.org/10.1080/01441647.2011.621557. that implementing regular communication and marketing efforts in Chen, C., Achtari, G., Majkut, K., Sheu, J.B., 2017. Balancing equity and cost in rural transportation management with multi-objective utility analysis and data rural municipalities as well as in regional centers during the service envelopment analysis: A case of Quinte West. Transp. Res. Part A Policy Pract. 95, life of a ridesharing scheme is likely to increase the participation of 148–165. https://doi.org/10.1016/j.tra.2016.10.015. local people and potential drivers in ridesharing. Another interesting Clewlow, R.R., Mishra, G.S., 2017. Disruptive Transportation: The Adoption, Utilization, and Impacts of Ride-Hailing in the United States. Institute of Transportation Studies. subject for future research is the impact of built environment on Universitiy of California. ridesharing demand. We hypothesize that in a rural context, tradi- Contreras, S.D., Paz, A., 2018. The effects of ride-hailing companies on the taxicab tional built environment factors such as population and employment industry in Las Vegas, Nevada. Transp. Res. Part A Policy Pract. 115, 63–70. https:// ‐ doi.org/10.1016/j.tra.2017.11.008. densities and land use diversity have no impacts on the demand of Docherty, I., Marsden, G., Anable, J., 2018. The governance of smart mobility. Transp. ridesharing whereas the location of regional public transport hubs Res. Part A Policy Pract. 115, 114–125. https://doi.org/10.1016/j.tra.2017.09.012. do have an impact. FOSD (Federal Office for Spatial Development), 2020. Web-GIS ARE. https://map.geo. The main purpose of the mixed methods approach was used as admin.ch/?lang=de&topic=are&catalogNodes=954,959,965&bgLayer=ch. swisstopo.pixelkarte-grau. complementarity to increase the scope of the inquiry, i.e., practice FOSD (Federal Office for Spatial Development), 2017. Verkehrsmodellierung im UVEK and outcomes of the integration of ridesharing with public transport (Traffic modulation in the DETEC - own translation). https://doi.org/10.5281/ systems, however, quantitative findings on the use of Taxito’s rideshar- zenodo.3251588. ’ Franco, P., Johnston, R., McCormick, E., 2019. Demand responsive transport: ing support results from qualitative interviews that Taxito s rideshar- Generation of activity patterns from mobile phone network data to support the ing scheme is a complement to public transport. This can be seen in operation of new mobility services. Transp. Res. Part A Policy Pract. 131, 244–266. a moderate number of ridesharing trips and more demand for rideshar- https://doi.org/10.1016/j.tra.2019.09.038. FSO (Federal Statistic Office), 2017. Mikrozensus Mobilität und Verkehr 2015 ing at weekends and late evenings when local bus operators reduce (Microcensus Mobility and Transport 2015 - Own translation). their services. FSO (Federal Statistic Office), 2014. Areas with urban character 2012: Explanatory report. Studies on transportation policies are often in favor of quantita- FSO (Federal Statistical Office), 2018a. Statpop (Version 1.10) [Dataset]. https://www. ‐ bfs.admin.ch/bfs/en/home/services/geostat/swiss-federal-statistics-geodata.html. tive methods such as modelling and macro trends (Marsden and FSO (Federal Statistical Office), 2018b. Statent (Version 1.4) [Dataset]. https://www. Reardon, 2017). However, new mobility solutions can involve a com- bfs.admin.ch/bfs/en/home/services/geostat/swiss-federal-statistics-geodata.html. plexity of policy making in realities as shown in this present case of Gray, D., Farrington, J., Shaw, J., Martin, S., Roberts, D., 2001. Car dependence in rural Scotland: Transport policy, devolution and the impact of the fuel duty escalator. J. the integration of ridesharing services with public transport systems. Rural Stud. 17, 113–125. https://doi.org/10.1016/S0743-0167(00)00035-8. A mixed methods approach therefore can grasp the complexity of Hall, J.D., Palsson, C., Price, J., 2018. Is Uber a substitute or complement for public how transportation policies get made and implemented at the local transit? J. Urban Econ. 108, 36–50. https://doi.org/10.1016/j.jue.2018.09.003. level. Henao, A., Marshall, W.E., 2018. The impact of ride-hailing on vehicle miles traveled. Transportation (Amst). 1–22. https://doi.org/10.1007/s11116-018-9923-2. The findings of this paper are based on an exploratory study. The Hofmann, D., Daskalakis, M., 2019. Ridesharing im ländlichen Raum [Ridesharing in impact between factors of the built environment and ridesharing use rural areas]. In: Daskalakis, M., Sommer, C., Rossnagel, A., Kepper, J. (Eds.), has been investigated purely by means of a spatial analysis. Regres- Ländliche Mobiliät Vernetzen: Ridesharing Im Ländlichen Raum Und Dessen fi Integration in Den Öffentlichen Nahverkehr [Connecting Rural Mobility: sion models to con rm this impact could not be carried out due to Ridesharing in Rural Areas and Its Integration in Regional Public Transportation]. the low level of observation of notice boards and their corresponding OEKOM, Munich, pp. 40–50. destinations. In addition, only three qualitative interviews were con- Joseph, R., 2018. Ride-Sharing Services: The Tumultuous Tale of the Rural Urban Divide, in: Twenty-Fourth Americas Conference on Information Systems. New Orleans. ducted with the involved stakeholders. Hence, any generalizations Kong, H., Zhang, X., Zhao, J., 2020. How does ridesourcing substitute for public transit? from our findings should take these data limitations into considera- A geospatial perspective in Chengdu. China. J. Transp. Geogr. 86,. https://doi.org/ tion. Nevertheless, the case‐study approach described here can pro- 10.1016/j.jtrangeo.2020.102769 102769. fi Krämer, P., Weiss, A., 2019. Vernetzung aller Mobilitätsangebote: der Odenwald in der vide insightful ndings into the integration of ridesharing with digitalen Welt der “On-Demand-Mobilität” [Connections of all mobility services: public transport in rural contexts, a still rather neglected area of Odenwald in the digital world of “on-demand mobility”]. Der Nahverkehr 1+2, research. 52–57. Marsden, G., Reardon, L., 2017. Questions of governance: Rethinking the study of transportation policy. Transp. Res. Part A Policy Pract. 101, 238–251. https://doi. CRediT authorship contribution statement org/10.1016/j.tra.2017.05.008. Malalgoda, N., Lim, S.H., 2019. Do transportation network companies reduce public transit use in the U.S.?. Transp. Res. Part A Policy Pract. 130, 351–372. https://doi. Vu Thi Thao: Conceptualization, Formal analysis, Writing ‐ origi- org/10.1016/j.tra.2019.09.051. nal draft, Writing ‐ review & editing. Sebastian Imhof: Methodology, Feigon, S., Murphy, C., 2016. Shared mobility and the transformation of public transit. ‐ Transit Cooperative Research Program, research report 188. Chicago. Investigation, Visualization, Formal analysis, Writing original draft. Murray, G., Chase, M., Kim, E., McBrayer, M., 2012. Ridesharing as a Complement to Widar von Arx: Methodology, Investigation, Formal analysis, Writing Transit: A Synthesis of Transit Practice. Washington DC. ‐ original draft. Nutley, S., 2003. Indicators of transport and accessibility problems in rural Australia. J. Transp. Geogr. 11, 55–71. https://doi.org/10.1016/S0966-6923(02)00052-2. Openrouteservice, 2020. API interactive examples: Pois service. https:// References openrouteservice.org/dev/#/api-docs/pois/post. Petersen, T., 2016. Watching the Swiss: A network approach to rural and exurban public transport. Transp. Policy 52, 175–185. https://doi.org/10.1016/ PostAuto provides rural and regional bus services throughout Switzerland. j.tranpol.2016.07.012. The interviewee’s identity was concealed at the interviewee’s request. Petersen, T., 2012. Public Transport for Exurban Settlements. The University of From 17 March 2020 to 24 September 2020, the Taxito ridesharing scheme had to Melbourne. suspend its services due to prevention measures connected with Covid-19 pandemic. Reck, D.J., Axhausen, K.W., 2019. Ridesourcing for the first/last mile: How do transfer This was because the local authorities considered Taxito to be a public transport penalties impact travel time savings? International Scientific Conference on service. Mobility and Transport (mobil.TUM 2019) Conference Proceedings. https://doi. ASTRA, 2018. Medienmitteilungen: 5,8 Millionen Menschen in der Schweiz haben das org/10.3929/ethz-b-000342106. Auto-Permis (Press Release, 5.8 Millions people in Switzerland's own driving Salemink, K., Strijker, D., Bosworth, G., 2017. Rural development in the digital age: A licence). systematic literature review on unequal ICT availability, adoption, and use in rural Benz, H., Kepper, J., 2019. Das Untersuchungsbeispiel Mobilfalt [The Investigation areas. J. Rural Stud. 54, 360–371. https://doi.org/10.1016/j.jrurstud.2015.09.001. Example Mobilfalt]. In: Daskalakis, M., Sommer, C., Rossnagel, A., Kepper, J. (Eds.), SBB (Swiss Federal Railways), 2020. DiDok. https://opendata.swiss/de/dataset/didok Ländliche Mobiliät Vernetzen: Ridesharing Im Ländlichen Raum Und Dessen Shannon, C.E., 1948. A mathematical theory of communication. Bell Syst. Tech. J. 27, Integration in Den Öffentlichen Nahverkehr [Connecting Rural Mobility: 379–423. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x. Ridesharing in Rural Areas and Its Integration in Regional Public Transportation]. OEKOM, Munich, pp. 66–85.

11 V.T. Thao et al. Transportation Research Interdisciplinary Perspectives 10 (2021) 100340

Srinivasan, R., 2019. Disrupting jobs and lives, in: Beyond the Valley: How Innovators VVL, 2014. Taxito: Öffentlicher Individualverkehr. Technischer Bericht Phase 1. (Taxito around the World Are Overcoming Inequality. MIT Press, Cambridge, MA, pp. - Public individual traffic. Technical report phase 1- Own translation). 113–124. VVL, 2017. Schlussbericht Pilotphase 2015-2017 Taxito: öffentlicher Individualverkehr Stiglic, M., Agatz, N., Savelsbergh, M., Gradisar, M., 2018. Enhancing urban mobility: (Final Report of pilot phase 2015-2017: Taxito as public individual transport - Own Integrating ride-sharing and public transit. Comput. Oper. Res. 90, 12–21. https:// translation). doi.org/10.1016/j.cor.2017.08.016. Yu, H., Peng, Z.R., 2019. Exploring the spatial variation of ridesourcing demand and its Thao, V.T., Ohnmacht, T., 2019. The impact of the built environment on travel behavior: relationship to built environment and socioeconomic factors with the The Swiss experience based on two National Travel Surveys. Res. Transp. Bus. geographically weighted Poisson regression. J. Transp. Geogr. 75, 147–163. Manag. 100386. https://doi.org/10.1016/j.rtbm.2019.100386. https://doi.org/10.1016/j.jtrangeo.2019.01.004.

12